Assimilator1
Member
No, below 95% is, read the op it will tell you this.
At 117% it appears your CPU is being slightly throttled (underclocked), is your CPU running hot?
IMPORTANT Explaining the AMD Ryzen "Power Reporting Deviation" -metric in HWiNFO
- Thread starter The Stilt
- Start date
At 117% it appears your CPU is being slightly throttled (underclocked), is your CPU running hot?
(Correct me if i am wrong please) As long i have tested with this new measure, i tried using stock values in bios and my % was 100 after Cinebench stress stest, but when i swap them to my configuration (Ryzen 5 3600 with 40 Ghz speed clock and 0,174V offset core voltage) it shows me 110%. So i understand its normal, because we both are undervolting our cpu (Ryzen 5 can reach 4.2 Ghz by default) and thats the reason why we are not using all the power to make us waste into 100%.
Oh thank you, then i overvaluated the measure so much, i was really worried
No, in fact i am running 40Ghz as i said, so i am undervolting a little bit for better temperatures (Ryzen 5 3600 holds 42Ghz by default)
agentnathan009
Member
No, look at screenshots in OP. The issue is a mobo manufacturer will use VRM values the make the CPU think that their is less amps being used so CPU thinks it has more headroom than it actually has. The effect is higher temperatures due to increased voltage for a cpu being tricked into thinking it only has 50 amp VRM utilized so CPU thinks it has more VRM headroom to boost higher. The built in processor monitoring will prevent CPU from cooking itself using AMD boosting algorithms.
People need to stop worrying about this, it isn’t as bad as some have made it out to be.
Assimilator1
Member
You're being way too vague about this and you haven't said whether you're referring to THGs article or the op's post.
Whilst I agree that very quick CPU burn out (as per THGs headline) is very unlikely (unless the PRD is seriously under), a more modest but still under reporting figure will run the CPU hotter & will have more power running through it. Depending how much, this will shorten CPU life the same as overclocking to varying degrees. It might not be much, but it's there.
agentnathan009
Member
Tom’s Hardware made it sound like chipocalypse and it isn’t. Inte mobo manufacturers have been doing this sort of thing for years.Ryzen CPUs have a built in monitoring system that prevents it from damaging itself or being damaged while using auto boasting algorithms. It may not prevent damage if you manually overclock too hard. I have yet to encounter CPUs that ended up dying early deathsdue to extra voltage under normal use, only with extreme overclocking usage are you likely to burn out a CPU.
I mean, any of % values should be a problem if there is no overheat in the CPU, right? (>80 degrees celcius).
Assimilator1
Member
40 GHz? lol, that's some overclock you've got!
, anyway, seeing as you've underclocked it to 4.0 GHz that explains it .I don't know why you're quoting me when I understand what the PRD is about
, did you mean to reply to Torse?He was getting 117% because he is underclocking his CPU, so the reading makes sense knowing that.
And I see you were referring to THGs article, fair enough, just earlier on you didn't mention this & it looked like you were referring to the op in this thread.
Btw, 1 point you're missing where mbrds are under reporting PRD, the CPU will run hotter & maybe the sole cause of excessive temps (over 80C).
agentnathan009
Member
Ryzen chips are able to go up to 95C before throttling significantly. Same as Intel chips. Are you going to own that CPU for 10 or more years? Probably not, so you don’t need to worry about it as the auto boosting and Ryzen FIT chip health monitoring will prevent it from burning out significantly early.
Assimilator1
Member
95C is the absolute redline and I wouldn't be running the CPU that hot! At those temps life could be an issue if keeping it for even 5 yrs (which many people do).
Also, when you're running a CPU hotter, that will in itself cause it to use more power.
Also, when you're running a CPU hotter, that will in itself cause it to use more power.
Assimilator1
Member
Good figure
Agentnathan
Oh & btw, with the stock cooler & settings, my 3600 was throttling at ~90C, are you sure about that 95C figure?
Agentnathan
Oh & btw, with the stock cooler & settings, my 3600 was throttling at ~90C, are you sure about that 95C figure?
agentnathan009
Member
Running hotter won’t necessarily use more power, electrons flow more freely, to a point, when conductor is heated. Vacuum tubes operate because of heating causing electrons to flow.
I agree that 95C is hot, but not redline, maybe a few hundred revs below redline, chip manufacturers build in safety margins just like nearly all manufacturers. Will the life of the chip be shortened running at 95C 24/7? Yes, but not by as much as you might think. Remember, RDNA GPUs have 110C hotspots and they still run fine.
TH has inflated the story.
I encourage everyone who hasn't done so yet to read the OP (and a few other posts by me and The Stilt later) where everything was described in detail.
Yet again: posting results while not running at full load has absolutely no meaning.
I encourage everyone who hasn't done so yet to read the OP (and a few other posts by me and The Stilt later) where everything was described in detail.
Yet again: posting results while not running at full load has absolutely no meaning.
agentnathan009
Member
Good figure
Agentnathan
Oh & btw, with the stock cooler & settings, my 3600 was throttling at ~90C, are you sure about that 95C figure?
I’m still looking for AMD states max but this is what I found so far. Temperature effects when acPu throttles, it has likely been throttling before 95C, but you should see shutdown and significant throttling at 95C. Chip is attempting to safe itself and it does have a thermal margin above 95C in case there is a temperature spike while shutting down or throttling. Manufacturers don’t run product to the edge, we do that by exceeding safety margins whether in automotive or electronic tweaking.
Attachments
Assimilator1
Member
I'm glad I'm not the op! I'd be tearing my hair out with the number of people here who've posted & not bothered to read the op, or read it properly! About 1/2 I reckon
[update] So far I've only come across an AT user posting info about this, but it is backed up with data, but I know it was talked about in an AT article too......
Btw 95C is for the Ryzen 1 not 2, see here https://www.amd.com/en/products/cpu/amd-ryzen-5-2600x
Not seeing any max temps for the 3600 page though!
Agreed about safety margins btw.
I'd be tearing my hair out with the number of people here who've posted & not bothered to read the op, or read it properly! About 1/2 I reckon Re your 1st point, yes it does, as I recall it, high temp's means higher resistance, which means slightly more power usage. It was an AnandTech article where I 1st come across this, but atm I can't remember which one as it was at least a few years ago!
[update] So far I've only come across an AT user posting info about this, but it is backed up with data, but I know it was talked about in an AT article too......
Btw 95C is for the Ryzen 1 not 2, see here https://www.amd.com/en/products/cpu/amd-ryzen-5-2600x
Not seeing any max temps for the 3600 page though!
Agreed about safety margins btw.
Last edited:
agentnathan009
Member
It is for Ryzen 2 as well, see screenshot. No, the colder a conductor is, the greater its resistance to current flow. Like I said, heat lowers resistance up to a point. Vacuum tubes in tube amps have to be heated to a certain temp before the resistance drops enough for electrons flow. CPUs are similar. However, the resistance to electron flow probably comes into play at various higher temps with various material. At some point the resistance may increase when temps reach a certain level, but it will vary depending on the material.I'm glad I'm not the op!
Re your 1st point, yes it does, as I recall it, high temp's means higher resistance, which means slightly more power usage. It was an AnandTech article where I 1st come across this, but atm I can't remember which one as it was at least a few years ago!
Btw 95C is for the Ryzen 1 not 2, see here https://www.amd.com/en/products/cpu/amd-ryzen-5-2600x
Not seeing any max temps for the 3600 page though!
Agreed about safety margins btw.
Attachments
powar deviator
Member
maybe im dumb but I DONT GET IT ... With 150A setting (50% of the actual), YOU HAVE THE BEST FREQUENCY 4106.6MHz, THE LOWEST POWER CONSUMPTION 91.553W AND LOWEST and peak CPU temperature of 79°C. T ???? so all wins no negatives or im i missing something here ( yes there is the ethical question etc) btw not ny native language, sorry
Assimilator1
Member
Agentnathan
Sorry but that's wrong, resistance drops with temperature in most materials (their are a few exceptions), hence superconductors have to be super cold.
I don't know what's going on vac tubes, I suspect that's down to something else. CPUs are not similar.
Your screen shot is of a Zen 1 2600 (I incorrectly stated Ryzen 1, 2 earlier, when I should've said Zen 1, 2.) Their appear to be no max temps listed for Zen 2, Ryzen 3000 series.
Btw, I got one thing wrong about power usage going up with temperatures, apparently it's mostly down to static leakage not resistance increasing. Checkout that thread and look at the tests he's done.
Sorry but that's wrong, resistance drops with temperature in most materials (their are a few exceptions), hence superconductors have to be super cold.
I don't know what's going on vac tubes, I suspect that's down to something else. CPUs are not similar.
Your screen shot is of a Zen 1
2600 (I incorrectly stated Ryzen 1, 2 earlier, when I should've said Zen 1, 2.) Their appear to be no max temps listed for Zen 2, Ryzen 3000 series.Btw, I got one thing wrong about power usage going up with temperatures, apparently it's mostly down to static leakage not resistance increasing. Checkout that thread and look at the tests he's done
.Assimilator1
Member
You're misunderstanding what PRD is, it's what the CPU is being told the power usage is, not what the actual power is.
If you restricted the power to a CPU to 50% of it's max, then you'd get great temps & low power consumption but you would not get the 'best frequency'
If you are saying the PRD is reading 50% when running CinebenchR20 (you didn't say), then the power being used by the CPU is being massively under reported, and so it can 'take' far more power than it should be, giving unnecessarily higher temps & power usage, what VID is being shown?
79C is a bit on the high side, what cooler have you got?
Last edited:
agentnathan009
Member
Sorry, I was wrong. Vacuum tubes have insulation effect so heating overcomes that insulation effect allowing electrons to flow.Agentnathan
Sorry but that's wrong, resistance drops with temperature in most materials (their are a few exceptions), hence superconductors have to be super cold.
I don't know what's going on vac tubes, I suspect that's down to something else. CPUs are not similar.
Your screen shot is of a Zen 1
Btw, I got one thing wrong about power usage going up with temperatures, apparently it's mostly down to static leakage not resistance increasing. Checkout that thread and look at the tests he's done
Per Anandtech article about this subject the electrical leakage has been mitigated quite a bit so it isn’t as bad as it was years ago.
Nope, for Zen 2 it is the same max temp.
Attachments
Last edited: